Conversational Interfaces have become an integral part of our daily interactions as they play a crucial role in facilitating human-machine communication. However, designing and developing effective experiences requires a deep understanding of the various conversational situations that can arise during interactions. Like our human daily conversations, Conversational Interfaces might face different situations which will affect the effectiveness of the interaction. Or what is the same, the understanding of the communicative exchange can be affected by these mentioned situations.
In this article, we will explore some key conversational situations which are handled and solved out of the box with OpenQuestion.
Repetitions
Sometimes, users may struggle to understand the response provided by the Conversational Interface. This is a typical situation especially for voice interactions, as we tend to multitask while interacting and instead of listening, we simply hear the message. It has happened to me personally that while I am experiencing a problem, I need to interact with someone to request help. Also, if I am completing a complex process and need support, I tend to try to do everything at once. In situations like this it is very easy for me to need to ask for instructions to be repeated back to me.
In such cases, it is essential to offer a simple way for users to request a repetition of the information provided. For this we have incorporated a specific flow in OpenQuestion that will take care of the repetition possibility. This specific flow takes into consideration the flow stack and revisited output features, as it is crucial to consider the previous context , i.e., being able to manage what has been said in the previous seconds and if there is an ongoing process activated. Having this information considered during the design and development will assure its success. In essence, incorporating a repetition strategy that allows users to ask the application to restate the information ensures the fluidity of a voice-based interaction.
Silences
During conversational interactions, users may unintentionally provide empty inputs or fail to provide the required information. For example, in my case I had inadvertently muted the phone while I was having a conversation which led to an empty input received by the Conversational Interface. In other situations, I have not been heard correctly for various reasons: wearing a mask that affects the acoustics, using a low-quality microphone that has not picked up the sound, etc. This also affected not only a phone call interaction but even the conversation I had with the people around me, who had issues capturing my voice correctly.
It is crucial for Conversational Interfaces to handle these situations gracefully and guide users towards the desired outcome. This specific experience or flow design is included in OpenQuestion as well, where it is activated only when detecting empty inputs. This situation is handled so in case no sound has been identified we can lead users back to the initial objective of the interaction.
Nonsense
In some instances, users may express nonsensical or irrelevant inputs unintentionally, posing a challenge for Conversational Interfaces. Sometimes it can also happen that we produce vocalizations or sounds that do not provide concrete meaning. In my personal experience, this has happened to me while talking on the phone in noisy environments like public transportation or walking my dogs in the park . Being on the phone trying to maintain a conversation surrounded by traffic or barking sounds can be quite challenging.
It is essential to handle such scenarios intelligently and guide users back on track, following a similar strategy as mentioned before. This can be handled first by informing the user of the situation like “I haven’t quite understood what you’re saying” and second by indicating the next step or providing an alternative like “Would you mind repeating the last thing you said?”
Multilingualism
Although natural language is the basis of these types of interactions, we might encounter users without knowledge of a particular language. A speaker’s ability or skill in the language of the project might hinder the facilitation of the interaction. An old friend of mine who moved to Germany to study faced difficulties because German was not her first language. In spite of the fact that the learning process was still improving, the preference for speaking in English was still necessary when discussing specific topics for example calling the technical support of the telco provider.
You can easily choose the languages required for the project as Teneo supports many different languages and OpenQuestion can change between those to provide the user with the desired language. So, each conversational experience can seamlessly adapt to the needs of this specific conversation.
Offering help
In conversational interactions, users often require additional assistance. As a Conversational Interface, it is crucial to anticipate these scenarios and offer proactive follow-up help. For example, if a user asks a question, the system can provide relevant suggestions or ask if they need further assistance. It is a nice and polite strategy to make sure to continue the conversation, allowing the user to decide the next step but in a partially guided way.
In addition to these common situations, a project may require a more generic conversational matters set. Within Teneo you can make use of Conversational Modules. These encompass all the necessary understanding and interaction components required to maintain a fluent dialogue and provide a conversational and human-like experience to the end-user. They go beyond small talk or casual conversations, serving as the fundamental dialogue capabilities that conversational AI applications need to exhibit human-like behavior.
In conclusion
This article has shown that the ability to design and develop effective Conversational Interfaces must address various conversational situations. One of the key characteristics inherent in OpenQuestion is its ability to offer further assistance, manage repetition requests, handle empty inputs, handle nonsense inputs intelligently, and even offer a multilingual experience that transcends language barriers. The true potential of conversational AI can be demonstrated by businesses by effectively handling all of the situations mentioned, which OpenQuestion embraced and managed for them.